We were discussing the current state of protection for human research subjects that have been deidentified. Mark Rothstein believes we need to do more to protect them. I think that that’s a laudable goal, but without big-time systemic changes it ain’t gonna happen:
I believe that if we are being honest with ourselves, then we will admit that under the current system, we are left with an unpalatable choice. From a policy standpoint, it is difficult to imagine how implementing further protections for the deidentified would pass the cost–benefit test. Because as things stand, we are already failing the identified…
…As someone who sits on one of eight institutional review boards (IRBs) at a major American academic medical center, it seems to me that the solution is not to impose still more byzantine rules on an already overwhelmed research community and those charged with overseeing it. We are already up to our eyeballs in new protocols, renewals, amendments, deferrals, conflicts of interest to adjudicate, legally authorized representatives, departmental reviews, and the occasional adverse event. Research on deidentified subjects must already receive a formal exemption in order to avoid the scrutiny of “true” human subjects research. Imagine if IRBs were suddenly required to grant those protocols a full review as well. Without systemic reform, the institutional Rube Goldberg machine otherwise known as human subjects protection would grind to a halt.
Is there a magic bullet? Um…no. But there’s stuff we can try.
Harvard’s George Church and his colleagues have decided that deidentification is sufficiently tenuous that it should be avoided altogether; his Personal Genome Project (in which I participate) makes no serious attempt to deidentify subjects. That’s certainly a less cumbersome option, although it is not clear how broadly it might be applied. Electronic health records might also mitigate the problem somewhat. Elsewhere, investigators at Vanderbilt have developed a massive deidentified DNA biobank from which patients must opt out if they wish not to participate. As Brothers and Clayton (2010) note, early indications are that the Vanderbilt group can indeed safeguard its subjects and that this approach is preferable to patients. One can also imagine involving a third-party “honest broker” to manage deidentification, re-identification, and all flows of subject data.
As I said, without profound changes, I expect that at some point the human subjects machine will break in a calamitous and irreparable way. And no amount of five-hour IRB meetings, 30-page consent forms or indignant bioethicists will be able to save it.
If you are interested in PDFs of any of the articles I’ve linked to here, let me know in the comments.